1 Ragel 5.21 - May 9, 2007
2 ========================
3 -Bug fix: when fhold was used in scanner pattern actions which get executed
4 on the last character of the pattern (pattern matches which do not require
5 any lookahead), fhold was modifying p instead of tokend. This was fixed and
6 the patact.rl test was modified to cover the case.
7 -Fixed typos in the guide, improved the state action embedding operator
9 -Implemented a better solution than the pri hack for resolving the '-'
10 ambiguity: force a shortest match of term.
11 -Fixed bugs in the binary searching for condition keys in both the Ruby and
13 -Can now embed the negative sense of a condition. Added a language-
14 independent test case for this feature and the necessary transformation
16 -Added new condition embedding syntax:
17 expr inwhen cond - The transitions into the machine (starting transitions).
18 expr outwhen cond - The pending transitions out of the machine.
19 -The argument to the variable statement which affects the name of the current
20 state variable was changed from "curstate" to "cs" (the default name used
21 for the current state)
22 -Implemented the other variables names in the variable statement. Now all
23 variables (p, pe, cs, top, stack, act, tokstart, tokend) can be renamed.
24 -Parse errors in the intermediate XML file now cause the backend to exit
25 immediately rather then forge on. The recovery infrastructure isn't there
26 and segfaults are likely.
27 -When no input is given to the backend program, it should not print an error
28 message, it should just return a non-zero exit status. The assumption is
29 that the frontend printed an error.
30 -The version number is now included in the intermediate file. An error is
31 emitted if there is a mismatch.
32 -The alphabet type is now communicated from the frontend to the backend using
33 a one-word internal name instead of an array offset.
34 -The Ruby host language types had been just copied from Java. Reduced them to
35 two basic types: char and int, both signed with the usual C sizes.
37 Ragel 5.20 - Apr 7, 2007
38 ========================
39 -The cs variable is now always initialized, unless the "nocs" option is given
40 to the write init command. If there is no main machine, cs is initialized to
41 the entry point defined by the last machine instantiation.
42 -A number of fixes were made to the Ruby code generator.
43 -The frontend now scans ruby comments and regular expressions.
44 -A transformation for Ruby was added to the language-independent test suite.
45 The Ruby code generator passes on all the language-independent tests.
46 -A new Ruby test and two language-independent tests were added.
47 -Some portability fixes were made (Patches from Josef Goettgens and Aaron
49 -Fixed a make dependency bug which caused a problem for parallel building
50 (Patch from Jeremy Hinegardner).
52 Ragel 5.19 - Mar 14, 2007
53 =========================
54 -Added an import statement to ragel. This statement takes a literal string as
55 an argument, interprets it as a file name, then scrapes the file for
56 sequences of tokens that match the following forms. Tokens inside ragel
57 sections are ignored. An example is in test/import1.rl
61 "define" name lit_string
62 -Added an export mechanism which writes defines for single character machines
63 that have been tagged with the export keyword in their definition. Defines
64 are used for C, ints for D, Java and Ruby. Examples of the export feature
65 are in test/export*.rl.
66 -All machine instantiations are now always generated, even if they are not
67 referenced. In the backend, entry points for all instantiations are written
68 out alongside start, error and first final states.
69 -If the main machine is not present then do not emit an error. Generate the
70 machine without a start state and do not initialize cs in the write init
72 -Added an option -l to rlgen-cd which inhibits the writing of #line
74 -Added a new syntax for verbose embeddings. This adds parentheses:
76 Verbose embeddings without parentheses can make code difficult to read
77 because they force a space in the middle of an action embedding. There is a
78 tendency to associtate spaces with concatenation. Without syntax
79 highlighting to make it clear that the embedding type is a keyword, the
80 problem is especially bad. The danger is that a verbose embedding could be
81 read as an embedding of the keyword representing the empbedding type. With
82 parentheses, verbose embeddings read much more clearly.
83 -Conditions now have a forced order when more than one is executed on a
84 single character. Previously ordering relied on pointers, which caused
85 results to vary by compiler. Ordering is now done using conditon action
86 declaration order. This fixes the failure of cond4.rl which occured with
87 g++ 4.1 and other compiler versions.
88 -In the port from flex to ragel, the name separator :: in Ragel code was
90 -In the examples directory switched from rlcodegen to rlgen-cd. Silenced a
91 warning in statechart.rl.
92 -In the root makefile the distclean target was fixed. It was calling clean in
93 the subdirs. In docs, the clean target was not deleting the new manpages for
94 the rlgen-* programs. Fixed.
95 -Portability and other fixes from Josef Goettgens were applied.
96 -The @datadir@ and @mandir@ variables are made use of in doc/Makefile.in for
97 specifying where documentation should be installed. Patch from Marcus
100 Ragel 5.18 - Feb 13, 2007
101 =========================
102 -There is now a 100% correspondence between state id numbers in the
103 intermediate XML file, Graphviz dot files and generated code. This was
104 achieved by moving code which determines if the error state is necessary
105 into the frontend, and then assigning state numbers before writing out the
107 -Backened class structure was reorganized to make it easier to add new code
108 generators without having to also modify the existing code generators.
109 -The C and D code generation executable was changed to rlgen-cd.
110 -The Java code generation was split out into it's own exectuable (rlgen-java)
111 to allow it to freely diverge from the C/D-based code generation.
112 -The graphviz dot file generation was also split out to it's own executable
114 -The Ruby code generation patch from Victor Hugo Borja was added. This is
115 highly experimental code and is not yet completely functional. It is in the
116 executable rlgen-ruby.
117 -The problem with large state machine machines in Java was fixed. This
118 problem was discovered by Colin Fleming, who also contributed a patch.
119 Rather than specify arrays as comma-separated lists of literals, array
120 initialization is now done in a static function. This is the approach used
121 by the Java compiler. Unlike the compiler Ragel is careful split large
122 initilization functions.
123 -The manual was expanded and reorganized somewhat.
124 -Eliminated per-example directories in examples/.
125 -Made some fixes to the pullscan.rl example.
126 -In the frontend CR characters are now treated as whitespace.
127 -Updated to the latest aapl. This completely eliminates the shallowCopy
128 function. With that, a definitive memory leak is fixed.
129 -Control codes with escape sequences are now printable characters (-p
130 option). Also, the space character is now printed as SP.
131 -Fixed the null dereference and consequential segfault which occurred when
132 trying to create empty machines with [] and // and /a[]b/.
133 -Fixed the segfault which occured when a machine reference failed.
134 -Discontinuing ragel.spec. It is more appropriate for this to be written by
135 package maintenance developers.
137 Ragel 5.17 - Jan 28, 2007
138 =========================
139 -The scanners and parsers in both the frontend and backend programs were
140 completely rewritten using Ragel and Kelbt.
141 -The '%when condition' syntax was functioning like '$when condition'. This
143 -In the Vim syntax file fixes to the matching of embedding operators were
144 made. Also, improvements to the sync patterns were made.
145 -Added pullscan.rl to the examples directory. It is an example of doing
146 pull-based scanning. Also, xmlscan.rl in rlcodegen is a pull scanner.
147 -The introduction chapter of the manual was improved. The manually-drawn
148 figures for the examples were replaced with graphviz-drawn figures.
150 Ragel 5.16 - Nov 20, 2006
151 =========================
152 -Policy change: the fhold and fexec directives did not function correctly in
153 scanner pattern actions. In this context manipulations of p may be lost or
154 made invalid. In the previous version of Ragel they were banned because of
155 this. Instead of banning these directives they have been fixed. The fexec
156 and fhold directives now manipulate tokend, which is now always used to
157 update p when the action terminates.
159 Ragel 5.15 - Oct 31, 2006
160 =========================
161 -A language independent test harness was introduced. Test cases can be
162 written using a custom mini-language in the embedded actions. This
163 mini-language is then translated to C, D and Java when generating the
164 language-specific test cases.
165 -Several existing tests have been ported to the language-independent format
166 and a number of new language-independent test cases have been added.
167 -The state-based embedding operators which access states that are not the
168 start state and are not final (the 'middle' states) have changed. They
170 <@/ eof action into middle states
171 <@! error action into middle states
172 <@^ local error action into middle states
173 <@~ to-state action into middle states
174 <@* from-state action into middle states
176 <>/ eof action into middle states
177 <>! error action into middle states
178 <>^ local error action into middle states
179 <>~ to-state action into middle states
180 <>* from-state action into middle states
181 -The verbose form of embeddings using the <- operator have been removed.
182 This syntax was difficult to remember.
183 -A new verbose form of state-based embedding operators have been added.
184 These are like the symbol versions, except they replace the symbols:
186 with literal keywords:
188 -The following words have been promoted to keywords:
189 when eof err lerr to from
190 -The write statment now gets its own lexical scope in the scanner to ensure
191 that commands are passed through as is (not affected by keywords).
192 -Bug fix: in the code generation of fret in scanner actions the adjustment to
193 p that is needed in some cases (dependent on content of patterns) was not
195 -The fhold directive, which decrements p, cannot be permitted in the pattern
196 action of a scanner item because it will not behave consistently. At the end
197 of a pattern action p could be decremented, set to a new value or left
198 alone. This depends on the contents of the scanner's patterns. The user
199 cannot be expected to predict what will happen to p.
200 -Conditions in D code require a cast to the widec type when computing widec.
201 -Like Java, D code also needs if (true) branches for control flow in actions
202 in order to fool the unreachable code detector. This is now abstracted in
203 all code generators using the CTRL_FLOW() function.
204 -The NULL_ITEM value in java code should be -1. This is needed for
205 maintaining tokstart.
207 Ragel 5.14 - Oct 1, 2006
208 ========================
209 -Fixed the check for use of fcall in actions embedded within longest match
210 items. It was emitting an error if an item's longest-match action had an
211 fcall, which is allowed. This bug was introduced while fixing a segfault in
213 -A new minimization option was added: MinimizeMostOps (-l). This option
214 minimizes at every operation except on chains of expressions and chains of
215 terms (eg, union and concat). On these chains it minimizes only at the last
216 operation. This makes test cases with many states compile faster, without
217 killing the performance on grammars like strings2.rl.
218 -The -l minimiziation option was made the default.
219 -Fixes to Java code: Use of the fc value did not work, now fixed. Static data
220 is now declared with the final keyword. Patch from Colin Fleming. Conditions
221 now work when generating Java code.
222 -The option -p was added to rlcodegen which causes printable characters to be
223 printed in GraphViz output. Patch from Colin Fleming.
224 -The "element" keyword no longer exists, removed from vim syntax file.
225 Updated keyword highlighting.
226 -The host language selection is now made in the frontend.
227 -Native host language types are now used when specifying the alphtype.
228 Previously all languages used the set defined by C, and these were mapped to
229 the appropriate type in the backend.
231 Ragel 5.13 - Sep 7, 2006
232 ========================
233 -Fixed a careless error which broke Java code generation.
235 Ragel 5.12 - Sep 7, 2006
236 ========================
237 -The -o flag did not work in combination with -V. This was fixed.
238 -The split code generation format uses only the required number of digits
239 when writing out the number in the file name of each part.
240 -The -T0, -F0 and -G0 codegens should write out the action list iteration
241 variables only when there are regular, to state or from state actions. The
242 code gens should not use anyActions().
243 -If two states have the same EOF actions, they are written out in the finish
245 -The split and in-place goto formats would sometimes generate _out when it is
246 not needed. This was fixed.
247 -Improved the basic partitioning in the split code gen. The last partition
248 would sometimes be empty. This was fixed.
249 -Use of 'fcall *' was not causing top to be initialized. Fixed.
250 -Implemented a Java backend, specified with -J. Only the table-based format
252 -Implemented range compression in the frontend. This has no effect on the
253 generated code, however it reduces the work of the backend and any programs
254 that read the intermediate format.
256 Ragel 5.11 - Aug 10, 2006
257 =========================
258 -Added a variable to the configure.in script which allows the building of
259 the parsers to be turned off (BUILD_PARSERS). Parser building is off by
260 default for released versions.
261 -Removed configure tests for bison defines header file. Use --defines=file
263 -Configure script doesn't test for bison, flex and gperf when building of the
264 parsers is turned off.
265 -Removed check for YYLTYPE structure from configure script. Since shipped
266 code will not build parsers by default, we don't need to be as accomodating
267 of other versions of bison.
268 -Added a missing include that showed up with g++ 2.95.3.
269 -Failed configure test for Objective-C compiler is now silent.
271 Ragel 5.10 - Jul 31, 2006
272 =========================
273 -Moved the check for error state higher in the table-based processing loop.
274 -Replaced naive implementations of condition searching with proper ones. In
275 the table-based formats the searching is also table-based. In the directly
276 executed formats the searching is also directly executable.
277 -The minimization process was made aware of conditions.
278 -A problem with the condition implementation was fixed. Previously we were
279 taking pointers to transitions and then using them after a call to
280 outTransCopy, which was a bad idea because they may be changed by the call.
281 -Added test mailbox3.rl which is based on mailbox2.rl but includes conditions
282 for restricting header and message body lengths.
283 -Eliminated the initial one-character backup of p just before resuming
285 -Added the -s option to the frontend for printing statistics. This currently
286 includes just the number of states.
287 -Sped up the generation of the in-place goto-driven (-G2) code style.
288 -Implemented a split version of in-place goto-driven code style. This code
289 generation style is suitable for producing fast implementations of very
290 large machines. Partitioning is currently naive. In the future a
291 high-quality partitioning program will be employed. The flag for accessing
292 this feature is -Pn, where n is the number of partitions.
293 -Converted mailbox1.rl, strings2.rl and cppscan1.rl tests to support the
294 split code generation.
295 -Fixes and updates were made to the runtests script: added -c for compiling
296 only, changed the -me option to -e, and added support for testing the split
299 Ragel 5.9 - Jul 19, 2006
300 ========================
301 -Fixed a bug in the include system which caused malformed output from the
302 frontend when the include was made from a multi-line machine spec and the
303 included file ended in a single line spec (or vice versa).
304 -Static data is now const.
305 -Actions which referenced states but were not embedded caused the frontend to
307 -Manual now built with pdflatex.
308 -The manual was reorganized and expanded. Chapter sequence is now:
309 Introduction, Constructing Machines, Embedding Actions, Controlling
310 Nondeterminism and Interfacing to the Host program.
312 Ragel 5.8 - Jun 17, 2006
313 ========================
314 -The internal representation of the alphabet type has been encapsulated
315 into a class and all operations on it have been defined as C++ operators.
316 -The condition implementation now supports range transitions. This allows
317 conditions to be embedded into arbitrary machines. Conditions are still
319 -More condition embedding operators were added
320 1. Isolate the start state and embed a condition into all transitions
323 2. Embed a condition into all transitions:
324 when cond OR $when cond OR $?cond
325 3. Embed a condition into pending out transitions:
327 -Improvements were made to the determinization process to support pending out
329 -The Vim sytax file was fixed so that :> doesn't cause the match of a label.
330 -The test suite was converted to a single-file format which uses less disk
331 space than the old directory-per-test format.
333 Ragel 5.7 - May 14, 2006
334 ========================
335 -Conditions will not be embedded like actions because they involve a
336 manipulation of the state machine they are specified in. They have therefore
337 been taken out of the verbose action embedding form (using the <- compound
338 symbol). A new syntax for specifying conditions has been created:
339 m = '\n' when {i==4};
340 -Fixed a bug which prevented state machine commands like fcurs, fcall, fret,
341 etc, from being accounted for in from-state actions and to-state actions.
342 This prevented some necessary support code from being generated.
343 -Implemented condition testing in remaining code generators.
344 -Configure script now checks for gperf, which is required for building.
345 -Added support for case-insensitive literal strings (in addition to regexes).
346 A case-insensitive string is made by appending an 'i' to the literal, as in
348 -Fixed a bug which caused all or expressions inside of all regular
349 expressions to be case-insensitive. For example /[fo]o bar/ would make the
350 [fo] part case-insensitive even though no 'i' was given following the
353 Ragel 5.6 - Apr 1, 2006
354 =======================
355 -Added a left-guarded concatenation operator. This operator <: is equivalent
356 to ( expr1 $1 . expr2 >0 ). It is useful if you want to prefix a sequence
357 with a sequence of a subset of the characters it matches. For example, one
358 can consume leading whitespace before tokenizing a sequence of whitespace
359 separated words: ( ' '* <: ( ' '+ | [a-z]+ )** )
360 -Removed context embedding code, which has been dead since 5.0.
362 Ragel 5.5 - Mar 28, 2006
363 ========================
364 -Implemented a case-insensitive option for regular expressions: /get/i.
365 -If no input file is given to the ragel program it reads from standard input.
366 -The label of the start state has been changed from START to IN to save on
367 required screen space.
368 -Bug fix: \0 was not working in literal strings, due to a change that reduced
369 memory usage by concatenating components of literal strings. Token data
370 length is now passed from the scanner to the paser so that we do not need to
371 rely on null termination.
373 Ragel 5.4 - Mar 12, 2006
374 ========================
375 -Eliminated the default transition from the frontend implementation. This
376 default transition was a space-saving optimization that at best could reduce
377 the number of allocated transitions by one half. Unfortunately it
378 complicated the implementation and this stood in the way of introducing
379 conditionals. The default transition may be reintroduced in the future.
380 -Added entry-guarded concatenation. This operator :>, is syntactic sugar
381 for expr1 $0 . expr >1. This operator terminates the matching of the first
382 machine when a first character of the second machine is matched. For
383 example in any* . ';' we never leave the any* machine. If we use any* :> ';'
384 then the any* machine is terminiated upon matching the semi-colon.
385 -Added finish-guarded concatenation. This operator :>>, is syntactic sugar
386 for expr1 $0 . expr @1. This operator is like entry guarded concatenation
387 except the first machine is terminated when the second machine enters a
388 final state. This is useful for delaying the guard until a full pattern is
389 matched. For example as in '/*' any* :>> '*/'.
390 -Added strong subtraction. Where regular subtraction removes from the first
391 machine any strings that are matched by the second machine, strong
392 subtraction removes any strings from the first that contain any strings of
393 the second as a substring. Strong subtraction is syntactic sugar for
394 expr1 - ( any* expr2 any* ).
395 -Eliminated the use of priorities from the examples. Replaced with
396 subtraction, guarded concatenation and longest-match kleene star.
397 -Did some initial work on supporting conditional transitions. Far from
398 complete and very buggy. This code will only be active when conditionals are
401 Ragel 5.3 - Jan 27, 2006
402 ========================
403 -Added missing semi-colons that cause the build to fail when using older
405 -Fix for D code: if the contents of an fexec is a single word, the generated
406 code will get interpreted as a C-style cast. Adding two brackets prevents
407 this. Can now turn eliminate the "access this.;" in cppscan5 that was used to
408 get around this problem.
409 -Improved some of the tag names in the intermediate format.
410 -Added unsigned long to the list of supported alphabet types.
411 -Added ids of actions and action lists to XML intermediate format. Makes it
413 -Updated to latest Aapl package.
415 Ragel 5.2 - Jan 6, 2006
416 ========================
417 -Ragel emits an error if the target of fentry, fcall, fgoto or fnext is inside
418 a longest match operator, or if an action embedding in a longest match
419 machine uses fcall. The fcall command can still be used in pattern actions.
420 -Made improvements to the clang, rlscan, awkemu and cppscan examples.
421 -Some fixes to generated label names: they should all be prefixed with _.
422 -A fix to the Vim syntax highlighting script was made
423 -Many fixes and updates to the documentation. All important features and
424 concepts are now documented. A second chapter describing Ragel's use
427 Ragel 5.1 - Dec 22, 2005
428 ========================
429 -Fixes to the matching of section delimiters in Vim syntax file.
430 -If there is a longest match machine, the tokend var is now initialized by
431 write init. This is not necessary for correct functionality, however
432 prevents compiler warnings.
433 -The rlscan example was ported to the longest match operator and changed to
435 -Fix to the error handling in the frontend: if there are errors in the lookup
436 of names at machine generation time then do not emit anything.
437 -If not compiling the full machine in the frontend (by using -M), avoid
438 errors and segfaults caused by names that are not part of the compiled
440 -Longest match bug fix: need to init tokstart when returing from fsm calls
441 that are inside longest match actions.
442 -In Graphviz drawing, the arrow into the start state is not a real
443 transition, do not draw to-state actions on the label.
444 -A bug fix to the handling of non-tag data within an XML tag was made.
445 -Backend exit value fixed: since the parser now accepts nothing so as to
446 avoid a redundant parse error when the frontend dies, we must force an
447 error. The backend should now be properly reporting errors.
448 -The longest match machine now has it's start state set final. An LM machine
449 is in a final state when it has not matched anything, when it has matched
450 and accepted a token and is ready for another, and when it has matched a
451 token but is waiting for some lookahead before determining what to do about
452 it (similar to kleene star).
453 -Element statement removed from some tests.
454 -Entry point names are propagated to the backend and used to label the entry
455 point arrows in Graphviz output.
457 Ragel 5.0 - Dec 17, 2005
458 ========================
459 (additional details in V5 release notes)
460 -Ragel has been split into two executables: A frontend which compiles
461 machines and emits them in an XML format, and a backend which generates code
462 or a Graphviz dot file from the XML input. The purpose of this split is to
463 allow Ragel to interface with other tools by means of the XML intermediate
464 format and to reduce complexity by strictly separating the previously
465 entangled phases. The intermediate format will provide a better platform
466 inspecting compiled machines and for extending Ragel to support other host
468 -The host language interface has been reduced significantly. Ragel no longer
469 expects the machine to be implemented as a structure or class and does not
470 generate functions corresponding to initialization, execution and EOF.
471 Instead, Ragel just generates the code of these components, allowing all of
472 them to be placed in a single function if desired. The user specifies a
473 machine in the usual manner, then indicates at which place in the program
474 text the state machine code is to be generated. This is done using the write
475 statement. It is possible to specify to Ragel how it should access the
476 variables it needs (such as the current state) using the access statement.
477 -The host language embedding delimiters have been changed. Single line
478 machines start with '%%' and end at newline. Multiline machines start with
479 '%%{' and end with '}%%'. The machine name is given with the machine
480 statement at the very beginning of the specification. This purpose of this
481 change is to make it easier separate Ragel code from the host language. This
482 will ease the addition of supported host languages.
483 -The structure and class parsing which was previously able to extract a
484 machine's name has been removed since this feature is dependent on the host
485 language and inhibits the move towards a more language-independent frontend.
486 -The init, element and interface statements have been made obsolete by the
487 new host language interface and have been removed.
488 -The fexec action statement has been changed to take only the new position to
489 move to. This statement is more useful for moving backwards and reparsing
490 input than for specifying a whole new buffer entirely and has been shifted
491 to this new use. Giving it only one argument also simplifies the parsing of
492 host code embedded in a Ragel specification. This will ease the addition of
493 supported host languages.
494 -Introduced the fbreak statement, which allows one to stop processing data
495 immediately. The machine ends up in the state that the current transition
496 was to go to. The current character is not changed.
497 -Introduced the noend option for writing the execute code. This inhibits
498 checking if we have reached pe. The machine will run until it goes into the
499 error state or fbreak is hit. This allows one to parse null-terminate
500 strings without first computing the length.
501 -The execute code now breaks out of the processing loop when it moves into
502 the error state. Previously it would run until pe was hit. Breaking out
503 makes the noend option useful when an error is encountered and allows
504 user code to determine where in the input the error occured. It also
505 eliminates needlessly iterating the input buffer.
506 -Introduced the noerror, nofinal and noprefix options for writing the machine
507 data. The first two inhibit the writing of the error state and the
508 first-final state should they not be needed. The noprefix eliminates the
509 prefixing of the data items with the machine name.
510 -Support for the D language has been added. This is specified in the backend
512 -Since the new host language interface has been reduced considerably, Ragel
513 no longer needs to distinguish between C-based languages. Support for C, C++
514 and Objective-C has been folded into one option in the backend: -C
515 -The code generator has been made independent of the languages that it
516 supports by pushing the language dependent apsects down into the lower
517 levels of the code generator.
518 -Many improvements to the longest match construction were made. It is no
519 longer considered experimental. A longest match machine must appear at the
520 top level of a machine instantiation. Since it does not generate a pure
521 state machine (it may need to backtrack), it cannot be used as an operand to
523 -References to the current character and current state are now completely
524 banned in EOF actions.
526 Ragel 4.2 - Sep 16, 2005
527 ========================
528 (additional details in V4 release notes)
529 -Fixed a bug in the longest match operator. In some states it's possible that
530 we either match a token or match nothing at all. In these states we need to
531 consult the LmSwitch on error so it must be prepared to execute an error
532 handler. We therefore need to init act to this error value (which is zero).
533 We can compute if we need to do this and the code generator emits the
534 initialization only if necessary.
535 -Changed the definition of the token end of longest match actions. It now
536 points to one past the last token. This makes computing the token length
537 easier because you don't have to add one. The longest match variables token
538 start, action identifier and token end are now properly initialized in
539 generated code. They don't need to be initialized in the user's code.
540 -Implemented to-state and from-state actions. These actions are executed on
541 transitions into the state (after the in transition's actions) and on
542 transitions out of the state (before the out transition's actions). See V4
543 release notes for more information.
544 -Since there are no longer any action embedding operators that embed both on
545 transitions and on EOF, any actions that exist in both places will be there
546 because the user has explicitly done so. Presuming this case is rare, and
547 with code duplication in the hands of the user, we therefore give the EOF
548 actions their own action switch in the finish() function. This is further
549 motivated by the fact that the best solution is to do the same for to-state
550 and from-state actions in the main loop.
551 -Longest match actions can now be specified using a named action. Since a
552 word following a longest match item conflicts with the concatenation of a
553 named machine, the => symbol must come immediately before a named action.
554 -The longest match operator permits action and machine definitions in the
555 middle of a longest match construction. These are parsed as if they came
556 before the machine definition they are contained in. Permitting action and
557 machine definitions in a longest match construction allows objects to be
558 defined closer to their use.
559 -The longest match operator can now handle longest match items with no
560 action, where previously Ragel segfaulted.
561 -Updated to Aapl post 2.12.
562 -Fixed a bug in epsilon transition name lookups. After doing a name lookup
563 the result was stored in the parse tree. This is wrong because if a machine
564 is used more than once, each time it may resolve to different targets,
565 however it will be stored in the same place. We now store name resolutions
566 in a separated data structure so that each walk of a parse tree uses the
567 name resolved during the corresponding walk in the name lookup pass.
568 -The operators used to embed context and actions into states have been
569 modified. The V4 release notes contain the full details.
570 -Added zlen builtin machine to represent the zero length machine. Eventually
571 the name "null" will be phased out in favour of zlen because it is unclear
572 whether null matches the zero length string or if it does not match any
573 string at all (as does the empty builtin).
574 -Added verbose versions of action, context and priority embedding. See the V4
575 release notes for the full details. A small example:
576 machine <- all exec { foo(); } <- final eof act1
577 -Bugfix for machines with epsilon ops, but no join operations. I had
578 wrongfully assumed that because epsilon ops can only increase connectivity,
579 that no states are ever merged and therefore a call to fillInStates() is not
580 necessary. In reality, epsilon transitions within one machine can induce the
581 merging of states. In the following, state 2 follows two paths on 'i':
582 main := 'h' -> i 'i h' i: 'i';
583 -Changed the license of the guide from a custom "do not propagate modified
584 versions of this document" license to the GPL.
586 Ragel 4.1 - Jun 26, 2005
587 ========================
588 (additional details in V4 release notes)
589 -A bug in include processing was fixed. Surrounding code in an include file
590 was being passed through to the output when it should be ignored. Includes
591 are only for including portions of another machine into he current. This
592 went unnoticed because all tested includes were wrapped in #ifndef ...
593 #endif directives and so did not affect the compilation of the file making
595 -Fixes were made to Vim syntax highlighting file.
596 -Duplicate actions are now removed from action lists.
597 -The character-level negation operator ^ was added. This operator produces a
598 machine that matches single characters that are not matched by the machine
599 it is applied to. This unary prefix operator has the same precedence level
601 -The use of + to specify the a positive literal number was discontinued.
602 -The parser now assigns the subtraction operator a higher precedence than
603 the negation of literal number.
605 Ragel 4.0 - May 26, 2005
606 ========================
607 (additional details in V4 release notes)
608 -Operators now strictly embed into a machine either on a specific class of
609 characters or on EOF, but never both. This gives a cleaner association
610 between the operators and the physical state machine entitites they operate
611 on. This change is made up of several parts:
612 1. '%' operator embeds only into leaving characters.
613 2. All global and local error operators only embed on error character
614 transitions, their action will not be triggerend on EOF in non-final
616 3. EOF action embedding operators have been added for all classes of states
617 to make up for functionality removed from other operators. These are
619 4. Start transition operator '>' no longer implicitly embeds into leaving
620 transtions when start state is final.
621 -Ragel now emits warnings about the improper use of statements and values in
622 action code that is embedded as an EOF action. Warnings are emitted for fpc,
623 fc, fexec, fbuf and fblen.
624 -Added a longest match construction operator |* machine opt-action; ... *|.
625 This is for repetition where an ability to revert to a shorter, previously
626 matched item is required. This is the same behaviour as flex and re2c. The
627 longest match operator is not a pure FSM construction, it introduces
628 transitions that implicitly hold the current character or reset execution to
629 a previous location in the input. Use of this operator requires the caller
630 of the machine to occasionally hold onto data after a call to the exectute
631 routine. Use of machines generated with this operator as the input to other
632 operators may have undefined results. See examples/cppscan for an example.
633 This is very experimental code.
634 -Action ids are only assigned to actions that are referenced in the final
635 constructed machine, preventing gaps in the action id sequence. Previously
636 an action id was assigned if the action was referenced during parsing.
637 -Machine specifications now begin with %% and are followed with an optional
638 name and either a single Ragel statement or a sequence of statements
640 -Ragel no longer generates the FSM's structure or class. It is up to the user
641 to declare the structure and to give it a variable named curs of type
642 integer. If the machine uses the call stack the user must also declare a
643 array of integers named stack and an integer variable named top.
644 -In the case of Objective-C, Ragel no longer generates the interface or
645 implementation directives, allowing the user to declare additional methods.
646 -If a machine specification does not have a name then Ragel tries to find a
647 name for it by first checking if the specification is inside a struct, class
648 or interface. If it is not then it uses the name of the previous machine
649 specification. If still no name is found then this is an error.
650 -Fsm specifications now persist in memory and statements accumulate.
651 -Ragel now has an include statement for including the statements of a machine
652 spec in another file (perhaps because it is the corresponding header file).
653 The include statement can also be used to draw in the statements of another
654 fsm spec in the current file.
655 -The fstack statement is now obsolete and has been removed.
656 -A new statement, simply 'interface;', indicates that ragel should generate
657 the machine's interface. If Ragel sees the main machine it generates the
658 code sections of the machine. Previously, the header portion was generated
659 if the (now removed) struct statement was found and code was generated if
660 any machine definition was found.
661 -Fixed a bug in the resolution of fsm name references in actions. The name
662 resolution code did not recurse into inline code items with children
663 (fgoto*, fcall*, fnext*, and fexec), causing a segfault at code generation
665 -Cleaned up the code generators. FsmCodeGen was made into a virtual base
666 class allowing for the language/output-style specific classes to inherit
667 both a language specific and style-specific base class while retaining only
668 one copy of FsmCodeGen. Language specific output can now be moved into the
669 language specific code generators, requiring less duplication of code in the
670 language/output-style specific leaf classes.
671 -Fixed bugs in fcall* implementation of IpgGoto code generation.
672 -If the element type has not been defined Ragel now uses a constant version
673 of the alphtype, not the exact alphtype. In most cases the data pointer of
674 the execute routine should be const. A non-const element type can still be
675 defined with the element statement.
676 -The fc special value now uses getkey for retrieving the current char rather
677 than *_p, which is wrong if the element type is a structure.
678 -User guide converted to TeX and updated for new 4.0 syntax and semantics.
680 Ragel 3.7 - Oct 31, 2004
681 ========================
682 -Bug fix: unreferenced machine instantiations causing segfault due to name
683 tree and parse tree walk becomming out of syncronization.
684 -Rewrote representation of inline code blocks using a tree data structure.
685 This allows special keywords such as fbuf to be used as the operatands of
687 -Documentation updates.
688 -When deciding whether or not to generate machine instantiations, search the
689 entire name tree beneath the instantiation for references, not just the
691 -Removed stray ';' in keller2.rl
692 -Added fexec for restarting the machine with new buffer data (state stays the
693 same), fbuf for retrieving the the start of the buf, and fblen for
694 retrieving the orig buffer length.
695 -Implemented test/cppscan2 using fexec. This allows token emitting and restart
696 to stay inside the execute routine, instead of leaving and re-entering on
698 -Changed examples/cppscan to use fexec and thereby go much faster.
699 -Implemented flex and re2c versions of examples/cppscan. Ragel version
700 goes faster than flex version but not as fast as re2c version.
701 -Merged in Objective-C patch from Erich Ocean.
702 -Turned off syncing with stdio in C++ tests to make them go faster.
703 -Renamed C++ code generaion classes with the Cpp Prefix instead of CC to make
705 -In the finish function emit fbuf as 0 cast to a pointer to the element type
706 so it's type is not interpreted as an integer.
707 -The number -128 underflows char alphabets on some architectures. Removed
709 -Disabled the keller2 test because it causes problems on many architectures
710 due to its large size and compilation requirements.
712 Ragel 3.6 - Jul 10, 2004
713 ========================
714 -Many documentation updates.
715 -When resolving names, return a set of values so that a reference in an
716 action block that is embedded more than once won't report distinct entry
717 points that are actually the same.
718 -Implemented flat tables. Stores a linear array of indicies into the
719 transition array and only a low and high key value. Faster than binary
720 searching for keys but not usable for large alphabets.
721 -Fixed bug in deleting of transitions leftover from converstion from bst to
722 list implementation of transitions. Other code cleanup.
723 -In table based output calculate the cost of using an index. Don't use if
725 -Changed fstate() value available in init and action code to to fentry() to
726 reflect the fact that the values returned are intended to be used as targets
727 in fgoto, fnext and fcall statements. The returned state is not a unique
728 state representing the label. There can be any number of states representing
730 -Added keller2 test, C++ scanning tests and C++ scanning example.
731 -In table based output split up transitions into targets and actions. This
732 allows actions to be omitted.
733 -Broke the components of the state array into separate arrays. Requires
734 adding some fields where they could previously be omitted, however allows
735 finer grained control over the sizes of items and an overal size reduction.
736 Also means that state numbers are not an offset into the state array but
737 instead a sequence of numbers, meaning the context array does not have any
739 -Action lists and transition also have their types chosen to be the smallest
740 possible for accomodating the contained values.
741 -Changed curs state stored in fsm struct from _cs to curs. Keep fsm->curs ==
742 -1 while in machine. Added tests curs1 and curs2.
743 -Implemented the notion of context. Context can be embedded in states using
744 >:, $:, @: and %: operators. These embed a named context into start states,
745 all states, non-start/non-final and final states. If the context is declared
746 using a context statment
748 then the context can be quered for any state using fsm_name_ctx_name(state)
749 in C code and fsm_name::ctx_name(state) in C++ code. This feature makes it
750 possible to determine what "part" of the machine is currently active.
751 -Fixed crash on machine generation of graphs with no final state. If there
752 is no reference to a final state in a join operation, don't generate one.
753 -Updated Vim sytax: added labels to inline code, added various C++ keywords.
754 Don't highlight name separations as labels. Added switch labels, improved
755 alphtype, element and getkey.
756 -Fixed line info in error reporting of bad epsilon trans.
757 -Fixed fstate() for tab code gen.
758 -Removed references to malloc.h.
760 Ragel 3.5 - May 29, 2004
761 ========================
762 -When parse errors occur, the partially generated output file is deleted and
763 an non-zero exit status is returned.
764 -Updated Vim syntax file.
765 -Implemented the setting of the element type that is passed to the execute
766 routine as well as method for specifying how ragel should retrive the key
767 from the element type. This lets ragel process arbitrary structures inside
768 of which is the key that is parsed.
769 element struct Element;
770 getkey fpc->character;
771 -The current state is now implemented with an int across all machines. This
772 simplifies working with current state variables. For example this allows a
773 call stack to be implemented in user code.
774 -Implemented a method for retrieving the current state, the target state, and
776 fcurs -retrieve the current state
777 ftargs -retrieve the target state
778 fstate(name) -retrieve a named state.
779 -Implemented a mechanism for jumping to and calling to a state stored in a
781 fgoto *<expr>; -goto the state returned by the C/C++ expression.
782 fcall *<expr>; -call the state returned by the C/C++ expression.
783 -Implemented a mechanism for specifying the next state without immediately
784 transfering control there (any code following statement is executed).
785 fnext label; -set the state pointed to by label as the next state.
786 fnext *<expr>; -set the state returned by the C/C++ expression as the
788 -Action references are determined from the final machine instead of during
789 the parse tree walk. Some actions can be referenced in the parse tree but not
790 show up in the final machine. Machine analysis is now done based on this new
792 -Named state lookup now employs a breadth-first search in the lookup and
793 allows the user to fully qualify names, making it possible to specify
794 jumps/calls into parts of the machine deep in the name hierarchy. Each part
795 of name (separated by ::) employs a breadth first search from it's starting
797 -Name references now must always refer to a single state. Since references to
798 multiple states is not normally intended, it no longer happens
799 automatically. This frees the programmer from thinking about whether or not
800 a state reference is unique. It also avoids the added complexity of
801 determining when to merge the targets of multiple references. The effect of
802 references to multiple states can be explicitly created using the join
803 operator and epsilon transitions.
804 -M option was split into -S and -M. -S specifies the machine spec to generate
805 for graphviz output and dumping. -M specifies the machine definition or
807 -Machine function parameters are now prefixed with and underscore to
808 avoid the hiding of class members.
810 Ragel 3.4 - May 8, 2004
811 =======================
812 -Added the longest match kleene star operator **, which is synonymous
813 with ( ( <machine> ) $0 %1 ) *.
814 -Epsilon operators distinguish between leaving transitions (going to an
815 another expression in a comma separated list) and non-leaving transitions.
816 Leaving actions and priorities are appropriately transferred.
817 -Relative priority of following ops changed to:
821 If label is done first then the isolation of the start state in > operators
822 will cause the label to point to the old start state that doesn't have the
824 -Merged >! and >~, @! and @~, %! and %~, and $! and $~ operators to have one
825 set of global error action operators (>!, @!, %! and $!) that are invoked on
826 error by unexpected characters as well as by unexepected EOF.
827 -Added the fpc keyword for use in action code. This is a pointer to the
828 current character. *fpc == fc. If an action is invoked on EOF then fpc == 0.
829 -Added >^, @^, %^, and $^ local error operators. Global error operators (>!,
830 @!, $!, and %!) cause actions to be invoked if the final machine fails.
831 Local error actions cause actions to be invoked if if the current machine
833 -Changed error operators to mean embed global/local error actions in:
834 >! and !^ -the start state.
835 @! and @^ -states that are not the start state and are not final.
836 %! and %^ -final states.
837 $! and $^ -all states.
838 -Added >@! which is synonymous >! then @!
839 -Added >@^ which is synonymous >^ then @^
840 -Added @%! which is synonymous @! then %!
841 -Added @%^ which is synonymous >^ then @^
842 -FsmGraph representation of transition lists was changed from a mapping of
843 alphabet key -> transition objects using a BST to simply a list of
844 transition objects. Since the transitions are no longer divided by
845 single/range, the fast finding of transition objects by key is no longer
846 required functionality and can be eliminated. This new implementation uses
847 the same amount of memory however causes less allocations. It also make more
848 sense for supporting error transitions with actions. Previously an error
849 transition was represented by a null value in the BST.
850 -Regular expression ranges are checked to ensure that lower <= upper.
851 -Added printf-like example.
852 -Added atoi2, erract2, and gotcallret to the test suite.
853 -Improved build test to support make -jN and simplified the compiling and
856 Ragel 3.3 - Mar 7, 2004
857 =======================
858 -Portability bug fixes were made. Minimum and maximum integer values are
859 now taken from the system. An alignment problem on 64bit systems
862 Ragel 3.2 - Feb 28, 2004
863 ========================
864 -Added a Vim syntax file.
865 -Eliminated length var from generated execute code in favour of an end
866 pointer. Using length requires two variables be read and written. Using an
867 end pointer requires one variable read and written and one read. Results in
868 more optimizable code.
869 -Minimization is now on by default.
870 -States are ordered in output by depth first search.
871 -Bug in minimization fixed. States were not being distinguished based on
873 -Added null and empty builtin machines.
874 -Added EOF error action operators. These are >~, >@, $~, and %~. EOF error
875 operators embed actions to take if the EOF is seen and interpreted as an
876 error. The operators correspond to the following states:
878 -any state with a transition to a final state
879 -any state with a transiion out
881 -Fixed bug in generation of unreference machine vars using -M. Unreferenced
882 vars don't have a name tree built underneath when starting from
883 instantiations. Need to instead build the name tree starting at the var.
884 -Calls, returns, holds and references to fc in out action code are now
885 handled for ipgoto output.
886 -Only actions referenced by an instantiated machine expression are put into
887 the action index and written out.
888 -Added rlscan, an example that lexes Ragel input.
890 Ragel 3.1 - Feb 18, 2004
891 ========================
892 -Duplicates in OR literals are removed and no longer cause an assertion
894 -Duplicate entry points used in goto and call statements are made into
895 deterministic entry points.
896 -Base FsmGraph code moved from aapl into ragel, as an increasing amount
897 of specialization is required. Too much time was spent attempting to
898 keep it as a general purpose template.
899 -FsmGraph code de-templatized and heirarchy squashed to a single class.
900 -Single transitions taken out of FsmGraph code. In the machine construction
901 stage, transitions are now implemented only with ranges and default
902 transtions. This reduces memory consumption, simplifies code and prevents
903 covered transitions. However it requires the automated selection of single
904 transitions to keep goto-driven code lean.
905 -Machine reduction completely rewritten to be in-place. As duplicate
906 transitions and actions are found and the machine is converted to a format
907 suitable for writing as C code or as GraphViz input, the memory allocated
908 for states and transitions is reused, instead of newly allocated.
909 -New reduction code consolodates ranges, selects a default transition, and
910 selects single transitions with the goal of joining ranges that are split by
911 any number of single characters.
912 -Line directive changed from "# <num> <file>" to the more common format
913 "#line <num> <file>".
914 -Operator :! changed to @!. This should have happened in last release.
915 -Added params example.
917 Ragel 3.0 - Jan 22, 2004
918 ========================
919 -Ragel now parses the contents of struct statements and action code.
920 -The keyword fc replaces the use of *p to reference the current character in
922 -Machine instantiations other than main are allowed.
923 -Call, jump and return statements are now available in action code. This
924 facility makes it possible to jump to an error handling machine, call a
925 sub-machine for parsing a field or to follow paths through a machine as
926 determined by arbitrary C code.
927 -Added labels to the language. Labels can be used anywhere in a machine
928 expression to define an entry point. Also references to machine definitions
929 cause the implicit creation of a label.
930 -Added epsilon transitions to the language. Epsilon operators may reference
931 labels in the current name scope resolved when join operators are evaluated
932 and at the root of the expression tree of machine assignment/instantiation.
933 -Added the comma operator, which joins machines together without drawing any
934 transitions between them. This operator is useful in combination with
935 labels, the epsilon operator and user code transitions for defining machines
936 using the named state and transition list paradigm. It is also useful for
937 invoking transitions based on some analysis of the input or on the
939 -Added >!, :!, $!, %! operators for specifying actions to take should the
940 machine fail. These operators embed actions to execute if the machine
943 -any state with a transition to a final state
944 -any state with a transiion out
946 The general rule is that if an action embedding operator embeds an action
947 into a set of transitions T, then the error-counterpart with a ! embeds an
948 action into the error transition taken when any transition T is a candidate,
949 but does not match the input.
950 -The finishing augmentation operator ':' has been changed to '@'. This
951 frees the ':' symbol for machine labels and avoids hacks to the parser to
952 allow the use of ':' for both labels and finishing augmentations. The best
953 hack required that label names be distinct from machine definition names as
954 in main := word : word; This restriction is not good because labels are
955 local to the machine that they are used in whereas machine names are global
956 entities. Label name choices should not be restricted by the set of names
957 that are in use for machines.
958 -Named priority syntax now requires parenthesis surrounding the name and
959 value pair. This avoids grammar ambiguities now that the ',' operator has
960 been introduced and makes it more clear that the name and value are an
962 -Backslashes are escaped in line directive paths.
964 Ragel 2.2 - Oct 6, 2003
965 =======================
966 -Added {n}, {,n}, {n,} {n,m} repetition operators.
967 <expr> {n} -- exactly n repetitions
968 <expr> {,n} -- zero to n repetitions
969 <expr> {n,} -- n or more repetitions
970 <expr> {n,m} -- n to m repetitions
971 -Bug in binary search table in Aapl fixed. Fixes crashing on machines that
972 add to action tables that are implicitly shared among transitions.
973 -Tests using obsolete minimization algorithms are no longer built and run by
975 -Added atoi and concurrent from examples to the test suite.
977 Ragel 2.1 - Sep 22, 2003
978 ========================
979 -Bug in priority comparison code fixed. Segfaulted on some input with many
981 -Added two new examples.
983 Ragel 2.0 - Sep 7, 2003
984 =======================
985 -Optional (?), One or More (+) and Kleene Star (*) operators changed from
986 prefix to postfix. Rationale is that postfix version is far more common in
987 regular expression implementations and will be more readily understood.
988 -All priority values attached to transitions are now accompanied by a name.
989 Transitions no longer have default priority values of zero assigned
990 to them. Only transitions that have different priority values assigned
991 to the same name influence the NFA-DFA conversion. This scheme reduces
992 side-effects of priorities.
993 -Removed the %! statement for unsetting pending out priorities. With
994 named priorities, it is not necessary to clear the priorities of a
995 machine with $0 %! because non-colliding names can be used to avoid
997 -Removed the clear keyword, which was for removing actions from a machine.
998 Not required functionality and it is non-intuitive to have a language
999 feature that undoes previous definitions.
1000 -Removed the ^ modifier to repetition and concatenation operators. This
1001 undocumented feature prevented out transitions and out priorities from being
1002 transfered from final states to transitions leaving machines. Not required
1003 functionality and complicates the language unnecessarily.
1004 -Keyword 'func' changed to 'action' as a part of the phasing out of the term
1005 'function' in favour of 'action'. Rationale is that the term 'function'
1006 implies that the code is called like a C function, which is not necessarily
1007 the case. The term 'action' is far more common in state machine compiler
1009 -Added the instantiation statement, which looks like a standard variable
1010 assignment except := is used instead of =. Instantiations go into the
1011 same graph dictionary as definitions. In the the future, instantiations
1012 will be used as the target for gotos and calls in action code.
1013 -The main graph should now be explicitly instantiated. If it is not,
1014 a warning is issued.
1015 -Or literal basic machines ([] outside of regular expressions) now support
1016 negation and ranges.
1017 -C and C++ interfaces lowercased. In the C interface an underscore now
1018 separates the fsm machine and the function name. Rationale is that lowercased
1019 library and generated routines are more common.
1021 int fsm_init( struct clang *fsm );
1022 int fsm_execute( struct clang *fsm, char *data, int dlen );
1023 int fsm_finish( struct clang *fsm );
1026 int fsm::execute( char *data, int dlen );
1028 -Init, execute and finish all return -1 if the machine is in the error state
1029 and can never accept, 0 if the machine is in a non-accepting state that has a
1030 path to a final state and 1 if the machine is in an accepting state.
1031 -Accept routine eliminated. Determining whether or not the machine accepts is
1032 done by examining the return value of the finish routine.
1033 -In C output, fsm structure is no longer a typedef, so referencing requires
1034 the struct keyword. This is to stay in line with C language conventions.
1035 -In C++ output, constructor is no longer written by ragel. As a consequence,
1036 init routine is not called automatically. Allows constructor to be supplied
1037 by user as well as the return value of init to be examined without calling it
1039 -Static start state and private structures are taken out of C++ classes.
1041 Ragel 1.5.4 - Jul 14, 2003
1042 ==========================
1043 -Workaround for building with bison 1.875, which produces an
1044 optimization that doesn't build with newer version gcc.
1046 Ragel 1.5.3 - Jul 10, 2003
1047 ==========================
1048 -Fixed building with versions of flex that recognize YY_NO_UNPUT.
1049 -Fixed version numbers in ragel.spec file.
1051 Ragel 1.5.2 - Jul 7, 2003
1052 =========================
1053 -Transition actions and out actions displayed in the graphviz output.
1054 -Transitions on negative numbers handled in graphviz output.
1055 -Warning generated when using bison 1.875 now squashed.
1057 Ragel 1.5.1 - Jun 21, 2003
1058 ==========================
1059 -Bugs fixed: Don't delete the output objects when writing to standard out.
1060 Copy mem into parser buffer with memcpy, not strcpy. Fixes buffer mem errror.
1061 -Fixes for compiling with Sun WorkShop 6 compilers.
1063 Ragel 1.5.0 - Jun 10, 2003
1064 ==========================
1065 -Line directives written to the output so that errors in the action code
1066 are properly reported in the ragel input file.
1067 -Simple graphviz dot file output format is supported. Shows states and
1068 transitions. Does not yet show actions.
1069 -Options -p and -f dropped in favour of -d output format.
1070 -Added option -M for specifying the machine to dump with -d or the graph to
1072 -Error recovery implemented.
1073 -Proper line and column number tracking implemented in the scanner.
1074 -All action/function code is now embedded in the main Execute routine. Avoids
1075 duplication of action code in the Finish routine and the need to call
1076 ExecFuncs which resulted in huge code bloat. Will also allow actions to
1077 modify cs when fsm goto, call and return is supported in action code.
1078 -Fsm spec can have no statements, nothing will be generated.
1079 -Bug fix: Don't accept ] as the opening of a .-. range a reg exp.
1080 -Regular expression or set ranges (ie /[0-9]/) are now handled by the parser
1081 and consequently must be well-formed. The following now generates a parser
1082 error: /[+-]/ and must be rewritten as /[+\-]/. Also fixes a bug whereby ]
1083 might be accepted as the opening of a .-. range causing /[0-9]-[0-9]/ to
1085 -\v, \f, and \r are now treated as whitespace in an fsm spec.
1087 Ragel 1.4.1 - Nov 19, 2002
1088 ==========================
1089 -Compile fixes. The last release (integer alphabets) was so exciting
1090 that usual portability checks got bypassed.
1092 Ragel 1.4.0 - Nov 19, 2002
1093 ==========================
1094 -Arbitrary integer alphabets are now fully supported! A new language
1096 'alphtype <type>' added for specifying the type of the alphabet. Default
1097 is 'char'. Possible alphabet types are:
1098 char, unsigned char, short, unsigned short, int, unsigned int
1099 -Literal machines specified in decimal format can now be negative when the
1100 alphabet is a signed type.
1101 -Literal machines (strings, decimal and hex) have their values checked for
1102 overflow/underflow against the size of the alphabet type.
1103 -Table driven and goto driven output redesigned to support ranges. Table
1104 driven uses a binary search for locating single characters and ranges. Goto
1105 driven uses a switch statement for single characters and nested if blocks for
1107 -Switch driven output removed due to a lack of consistent advantages. Most of
1108 the time the switch driven FSM is of no use because the goto FSM makes
1109 smaller and faster code. Under certain circumstances it can produce smaller
1110 code than a goto driven fsm and be almost as fast, but some sporadic case
1111 does not warrant maintaining it.
1112 -Many warnings changed to errors.
1113 -Added option -p for printing the final fsm before minimization. This lets
1114 priorities be seen. Priorties are all reset to 0 before minimization. The
1115 exiting option -f prints the final fsm after minimization.
1116 -Fixed a bug in the clang test and example that resulted in redundant actions
1119 Ragel 1.3.4 - Nov 6, 2002
1120 =========================
1121 -Fixes to Chapter 1 of the guide.
1122 -Brought back the examples and made them current.
1123 -MSVC is no longer supported for compiling windows binaries because its
1124 support for the C++ standard is frustratingly inadequate, it will cost money
1125 to upgrade if it ever gets better, and MinGW is a much better alternative.
1126 -The build system now supports the --host= option for building ragel
1127 for another system (used for cross compiling a windows binary with MinGW).
1128 -Various design changes and fixes towards the goal of arbitrary integer
1129 alphabets and the handling of larger state machines were made.
1130 -The new shared vector class is now used for action lists in transitions and
1131 states to reduce memory allocations.
1132 -An avl tree is now used for the reduction of transitions and functions of an
1133 fsm graph before making the final machine. The tree allows better scalability
1134 and performance by not requiring consecutively larger heap allocations.
1135 -Final stages in the separation of fsm graph code from action embedding and
1136 priority assignment is complete. Makes the base graph leaner and easier to reuse
1137 in other projects (like Keller).
1139 Ragel 1.3.3 - Oct 22, 2002
1140 ==========================
1141 -More diagrams were added to section 1.7.1 of the user guide.
1142 -FSM Graph code was reworked to spearate the regex/nfa/minimizaion graph
1143 algorithms from the manipulation of state and transition properties.
1144 -An rpm spec file from Cris Bailiff was added. This allows an rpm for ragel
1145 to be built with the command 'rpm -ta ragel-x.x.x.tar.gz'
1146 -Fixes to the build system and corresponding doc updates in the README.
1147 -Removed autil and included the one needed source file directly in the top
1148 level ragel directory.
1149 -Fixed a bug that nullified the 20 times speedup in large compilations
1150 claimed by the last version.
1151 -Removed awk from the doc build (it was added with the last release -- though
1152 not mentioned in the changelog).
1153 -Install of man page was moved to the doc dir. The install also installs the
1154 user guide to $(PREFIX)/share/doc/ragel/
1156 Ragel 1.3.2 - Oct 16, 2002
1157 ==========================
1158 -Added option -v (or --version) to show version information.
1159 -The subtract operator no longer removes transition data from the machine
1160 being subtracted. This is left up to the user for the purpose of making it
1161 possible to transfer transitions using subtract and also for speeding up the
1162 subtract routine. Note that it is possible to explicitly clear transition
1163 data before a doing a subtract.
1164 -Rather severe typo bug fixed. Bug was related to transitions with higher
1165 priorities taking precedence. A wrong ptr was being returned. It appears to
1166 have worked most of the time becuase the old ptr was deleted and the new one
1167 allocated immediatly after so the old ptr often pointed to the same space.
1169 -Bug in the removing of dead end paths was fixed. If the start state
1170 has in transitions then those paths were not followed when finding states to
1171 keep. Would result in non-dead end states being removed from the graph.
1172 -In lists and in ranges are no longer maintained as a bst with the key as the
1173 alphabet character and the value as a list of transitions coming in on that
1174 char. There is one list for each of inList, inRange and inDefault. Now that
1175 the required functionality of the graph is well known it is safe to remove
1176 these lists to gain in speed and footprint. They shouldn't be needed.
1177 -IsolateStartState() runs on modification of start data only if the start
1178 state is not already isolated, which is now possible with the new in list
1180 -Concat, Or and Star operators now use an approximation to
1181 removeUnreachableStates that does not require a traversal of the entire
1182 graph. This combined with an 'on-the-fly' management of final bits and final
1183 state status results is a dramatic speed increase when compiling machines
1184 that use those operators heavily. The strings2 test goes 20 times faster.
1185 -Before the final minimization, after all fsm operations are complete,
1186 priority data is reset which enables better minimization in cases where
1187 priorities would otherwise separate similar states.
1189 Ragel 1.3.1 - Oct 2, 2002
1190 =========================
1191 -Range transitions are now used to implement machines made with /[a-z]/ and
1192 the .. operator as well as most of the builtin machines. The ranges are not
1193 yet reflected in the output code, they are expanded as if they came from the
1194 regular single transitions. This is one step closer to arbitrary integer
1196 -The builtin machine 'any' was added. It is equiv to the builtin extend,
1197 matching any characters.
1198 -The builtin machine 'cntrl' now includes newline.
1199 -The builtin machine 'space' now includes newline.
1200 -The builtin machine 'ascii' is now the range 0-127, not all characters.
1201 -A man page was written.
1202 -A proper user guide was started. Chapter 1: Specifying Ragel Programs
1203 was written. It even has some diagrams :)
1205 Ragel 1.3.0 - Sep 4, 2002
1206 =========================
1207 -NULL keyword no longer used in table output.
1208 -Though not yet in use, underlying graph structure changed to support range
1209 transitions. As a result, most of the code that walks transition lists is now
1210 implemented with an iterator that hides the complexity of the transition
1211 lists and ranges. Range transitions will be used to implement /[a-z]/ style
1212 machines and machines made with the .. operator. Previously a single
1213 transition would be used for each char in the range, which is very costly.
1214 Ranges eliminate much of the space complexity and allow for the .. operator
1215 to be used with very large (integer) alphabets.
1216 -New minimization similar to Hopcroft's alg. It does not require n^2 space and
1217 runs close to O(n*log(n)) (an exact analysis of the alg is very hard). It is
1218 much better than the stable and approx minimization and obsoletes them both.
1219 An exact implementation of Hopcroft's alg is desirable but not possible
1220 because the ragel implementation does not assume a finite alphabet, which
1221 Hopcroft's requires. Ragel will support arbitrary integer alphabets which
1222 must be treated as an infinite set for implementation considerations.
1223 -New option -m using above described minimization to replace all previous
1224 minimization options. Old options sill work but are obsolete and not
1226 -Bug fixed in goto style output. The error exit set the current state to 0,
1227 which is actually a valid state. If the machine was entered again it would go
1228 into the first state, very wrong. If the first state happened to be final then
1229 an immediate finish would accept when in fact it should fail.
1230 -Slightly better fsm minimization now capable due to clearing of the
1231 transition ordering numbers just prior to minimization.
1233 Ragel 1.2.2 - May 25, 2002
1234 ==========================
1235 -Configuration option --prefix now works when installing.
1236 -cc file extension changed to cpp for better portability.
1237 -Unlink of output file upon error no longer happens, removes dependency on
1238 unlink system command.
1239 -All multiline strings removed: not standard c++.
1240 -Awk build dependency removed.
1241 -MSVC 6.0 added to the list of supported compilers (with some tweaking of
1242 bison and flex output).
1244 Ragel 1.2.1 - May 13, 2002
1245 ==========================
1246 -Automatic dependencies were fixed, they were not working correctly.
1247 -Updated AUTHORS file to reflect contributors.
1248 -Code is more C++ standards compliant: compiles with g++ 3.0
1249 -Fixed bugs that only showed up in g++ 3.0
1250 -Latest (unreleased) Aapl.
1251 -Configuration script bails out if bison++ is installed. Ragel will not
1252 compile with bison++ because it is coded in c++ and bison++ automatically
1253 generates a c++ parser. Ragel uses a c-style bison parser.
1255 Ragel 1.2.0 - May 3, 2002
1256 =========================
1257 -Underlying graph structure now supports default transitions. The result is
1258 that a transition does not need to be made for each char of the alphabet
1259 when making 'extend' or '/./' machines. Ragel compiles machines that
1260 use the aforementioned primitives WAY faster.
1261 -The ugly hacks needed to pick default transitions now go away due to
1262 the graph supporting default transitions directly.
1263 -If -e is given, but minimization is not turned on, print a warning.
1264 -Makefiles use automatic dependencies.
1266 Ragel 1.1.0 - Apr 15, 2002
1267 ==========================
1268 -Added goto fsm: much faster than any other fsm style.
1269 -Default operator (if two machines are side by side with no operator
1270 between them) is concatenation. First showed up in 1.0.4.
1271 -The fsm machine no longer auotmatically builds the flat table for
1272 transition indicies. Instead it keeps the key,ptr pair. In tabcodegen
1273 the flat table is produced. This way very large alphabets with sparse
1274 transitions will not consume large amounts of mem. This is also in prep
1275 for fsm graph getting a default transition.
1276 -Generated code contains a statement explicitly stating that ragel fsms
1277 are NOT covered by the GPL. Technically, Ragel copies part of itself
1278 to the output to make the generic fsm execution routine (for table driven
1279 fsms only) and so the output could be considered under the GPL. But this
1280 code is very trivial and could easlily be rewritten. The actual fsm data
1281 is subject to the copyright of the source. To promote the use of Ragel,
1282 a special exception is made for the part of the output copied from Ragel:
1283 it may be used without restriction.
1284 -Much more elegant code generation scheme is employed. Code generation
1285 class members need only put the 'codegen' keyword after their 'void' type
1286 in order to be automatically registerd to handle macros of the same name.
1287 An awk script recognises this keyword and generates an appropriate driver.
1288 -Ragel gets a test suite.
1289 -Postfunc and prefunc go away because they are not supported by non
1290 loop-driven fsms (goto, switch) and present duplicate functionality.
1291 Universal funcs can be implemented by using $ operator.
1292 -Automatic dependencies used in build system, no more make depend target.
1293 -Code generation section in docs.
1294 -Uses the latests aapl.
1296 Ragel 1.0.5 - Mar 3, 2002
1297 =========================
1298 -Bugfix in SetErrorState that caused an assertion failure when compiling
1299 simple machines that did not have full transition tables (and thus did
1300 not show up on any example machines). Assertion failure did not occur
1301 when using the switch statement code as ragel does not call SetErrorState
1303 -Fixed some missing includes, now compiles on redhat.
1304 -Moved the FsmMachTrans Compare class out of FsmMachTrans. Some compilers
1305 don't deal with nested classes in templates too well.
1306 -Removed old unused BASEREF in fsmgraph and ragel now compiles using
1307 egcs-2.91.66 and presumably SUNWspro. The baseref is no longer needed
1308 because states do not support being elements in multiple lists. I would
1309 rather be able to support more compilers than have this feature.
1310 -Started a README with compilation notes. Started an AUTHORS file.
1311 -Started the user documentation. Describes basic machines and operators.
1313 Ragel 1.0.4 - Mar 1, 2002
1314 =========================
1315 -Ported to the version of Aapl just after 2.2.0 release. See
1316 http://www.ragel.ca/aapl/ for details on aapl.
1317 -Fixed a bug in the clang example: the newline machine was not stared.
1318 -Added explanations to the clang and mailbox examples. This should
1319 help people that want to learn the lanuage as the manual is far from
1322 Ragel 1.0.3 - Feb 2, 2002
1323 =========================
1324 -Added aapl to the ragel tree. No longer requires you to download
1325 and build aapl separately. Should avoid discouraging impatient users
1326 from compiling ragel.
1327 -Added the examples to the ragel tree.
1328 -Added configure script checks for bison and flex.
1329 -Fixed makefile so as not to die with newer versions of bison that
1330 write the header of the parser to a .hh file.
1331 -Started ChangeLog file.
1333 Ragel 1.0.2 - Jan 30, 2002
1334 ==========================
1335 -Bug fix in calculating highIndex for table based code. Was using
1336 the length of out tranisition table rather than the value at the
1338 -If high/low index are at the limits, output a define in their place,
1339 not the high/low values themselves so as not to cause compiler warnings.
1340 -If the resulting machines don't have any indicies or functions, then
1341 omit the empty unrefereced static arrays so as not to cause compiler
1342 warnings about unused static vars.
1343 -Fixed variable sized indicies support. The header cannot have any
1344 reference to INDEX_TYPE as that info is not known at the time the header
1345 data is written. Forces us to use a void * for pointers to indicies. In
1346 the c++ versions we are forced to make much of the data non-member
1347 static data in the code portion for the same reason.
1349 Ragel 1.0.1 - Jan 28, 2002
1350 ==========================
1351 -Exe name change from reglang to ragel.
1352 -Added ftabcodegen output code style which uses a table for states and
1353 transitions but uses a switch statement for the function execution.
1354 -Reformatted options in usage dump to look better.
1355 -Support escape sequences in [] sections of regular expressions.
1357 Ragel 1.0 - Jan 25, 2002
1358 ========================